POLICY RECOMMENDATIONS FOR THE APPLICATIONS OF IMPACT … · 2016-07-21 · Policy Recommendations for the Applications of Impact Bonds: A Summary of Lessons Learned from the First

POLICY RECOMMENDATIONS FOR THE APPLICATIONS OF IMPACT BONDS

A summary of lessons learned from the first five years of experience worldwide

Emily Gustafsson-WrightSophie Gardiner

n o v e m b e r 2015

Policy Recommendations for the Applications of Impact Bonds: A Summary of Lessons Learned from the First Five Years of Experience Worldwide

Global Economy and Development Program at BROOKINGS02

The Brookings Institution is a nonprofit organization devoted to independent research and policy solutions. Its mission is to conduct high-quality, independent research and, based on that research, to provide innovative, practical recommendations for policymakers and the public. The conclusions and recommendations of any Brookings publication are solely those of its author(s), and do not reflect the views of the Institution, its management, or its other scholars.



The persistence and enormity of a number of global social challenges such as inadequate

access to quality health care and education re-quires diverse and innovative solutions that ad-dress the weaknesses of traditional approaches. The social impact bond (SIB), also called pay-for-success (PFS) in the United States and the so-cial benefit bond (SBB) in Australia, has garnered particular attention in recent years as a potential model to address some of these challenges. SIBs (and the related development impact bonds) are a mechanism that harnesses private capital for social services and encourages outcome achieve-ment by making repayment contingent upon suc-cess. In the SIB model, private investors put up capital to fund a social intervention and govern-ments repay the investor only if an agreed-upon outcome is achieved.1 The basic impact bond structure and mechanics are shown in Figure 1. An impact bond transaction typically involves four major types of actors, in addition to the population in need. Investors provide capital for a service provider to deliver social services to a popula-tion in need. The outcome funder (a government entity, or in the case of a development impact bond, a third party) agrees to repay the investors if pre-determined outcomes are achieved. The in-termediary can play multiple roles but often has

the responsibility of raising capital and bringing the stakeholders together to determine and agree upon the transactional details. In addition to these four players, an evaluator may be used to assess the outcomes.

Development impact bond (DIB) is a term used for a social impact bond that is implemented in low- and middle-income countries, where a donor agency or a foundation is the outcome funder as opposed to the government (although some combination of govern-ment with third party is also possible, in which case the term SIB might be preferable).2

Impact bonds have arisen out of three global trends in social service provision. First, over the past decade, there has been greater emphasis on evidence as a basis for financing, which has led to a movement to evaluate program impact in both the developed and the developing worlds. These impact evaluations have provided governments and development agencies with the information needed to increase investment in the most cost-ef-fective interventions,3 though more work is needed to ensure the findings are utilized. Second, there has been a greater focus on value for money, leading to the use of performance-based financing and output-based aid programs in both developed countries and by development finance institutions

1 The term “social impact bond” has also been used for issuance of traditional, fixed-yield bonds to raise capital for social programs. This differs from the definition of “social impact bond” used in this study, in that this study defines “social impact bonds” to be arrangements where payments to investors are dependent on and positively correlated with positive outcomes.

2 Center for Global Development and Social Finance (2013).3 See discussion of Mexico’s Progresa/Oportunidades on page 27 of Jones (2009).

I. A NEW APPROACH TO ADDRESS SOCIAL CHALLENGES: IMPACT BONDS





globally. There has been recognition that the pri-vate sector can complement the public sector in both financing and delivery of social services. This has resulted in an explosion in the participation of the private sector through, for example, public-pri-vate partnership models. Third, increased appetite for achieving double bottom line (social and finan-cial) returns has led to a growth in impact invest-ing globally. In 2014, in a survey of 125 investors,4 the global market was estimated to be valued at $46 billion, of which $32 billion was invested in developing countries.5 However, impact investing

still represents a relatively small fraction of total assets under management, which were estimated to be worth $64 trillion in 2012 and expected to exceed $100 trillion by 2020.6

Impact bonds bring together elements of these various strains of thinking and policy action into one instrument. SIBs are a form of public pri-vate partnership (PPP), which have traditionally been used to finance infrastructure in developed countries. SIBs differ from other results- or per-formance-based financing mechanisms in that

4 This includes only investors with assets under management of $10 million or above.5 Saltuk (2014).6 PricewaterhouseCoopers (2014).

INTERMEDIARY

EVALUATOR

SERVICE PROVIDER

POPULATIONIN NEED

OUTCOME FUNDER

6. Pay for Success

3. Deliver Services

4. Achieve Outcomes

5. Evaluate Impact

2. Coordinate, Structure Deal, & Manage Performance

1. Investment of Principal7. Return of Principalplus Interest

INVESTORS

Figure 1. Impact bond mechanics



financing is provided upfront, rather than when re-sults are attained, and results are usually related to outcomes as opposed to outputs. In addition, in contrast to performance-based programs such as Program for Results (P4R) or results-based fi-nancing (RBF) used by the World Bank, impact bonds bring in private sector rigor and perfor-mance management to drive results.7

Impact bonds are being used for different reasons in different locations, but thus far have been used to solve at least one of the following problems:

1. Lack of knowledge about the most effective intervention model;

2. Lack of political will to invest in a service;

3. Lack of upfront funding for a program that leads to later savings or value to society;

4. Lack of government or service provider ca-pacity to provide a necessary combination of services, manage services, or connect data across agencies.

Four basic criteria are necessary for an impact bond. First, a meaningful and measureable out-come must exist. Meaningful outcomes should be proxies for longer-term social and economic out-comes, which may or may not lead to fiscal sav-ings. Second, these outcomes must be achiev-able and measurable within a time horizon that is appealing to investors and outcome funders. Third, the evidence of service provider success in achieving outcomes must match the risk appetite of investors and outcome funders. Fourth, legal and political conditions must allow governments (in their role as outcome funders) to pay for out-comes rather than service inputs and make pay-ments beyond the fiscal year in which a contract is made and, for that matter, to pay for outcomes at all. If an intermediary party is managing the impact bond, it may also be necessary for legal condi-tions to support the ability of the government to

direct funds to an intermediary in a transaction and for the intermediary to have the authority to make certain decisions, such as selecting a service pro-vider. In addition to these criteria, there must be a set of actors who possess the expertise, will, and dedication required to carry out the transaction.

Though impact bonds have only existed for five years and much remains to be learned, some trends have emerged in the areas that are inap-propriate for the impact bond mechanism to be used. Impact bonds are not recommendable in the following cases: 1. Where government is willing to fund an out-

comes-based, proven intervention without ex-ternal support;

2. For funding core government services, such as the provision of the entirety of a primary or secondary education system (though pieces of it could be funded);

3. For interventions where outcomes are diffuse, impossible to measure, do not fully capture all areas of intervention impact, or where there are no proximate measures available within a timeframe that is reasonable to investors and outcome funders;

4. Where there is a risk of perverse incentives with a given outcome metric;

5. Where there is insufficient evidence of in-tervention impact for investor and outcome funder risk appetite;

6. Where stakeholders are unable to legally en-ter into the required contracts. For example, where outcome funders are unable to commit future funds to non-state actors;

7. Where it is impossible to ensure outcome funder ability to repay investors, as with a gov-ernment with a poor credit rating;

8. Where stakeholders are unable to establish necessary data sharing systems;

7 Burand (2013); Center for Global Development and Social Finance (2013); Bloomgarden, Eddy, & Levey (2014).



9. Where investors may unduly pressure service providers;

10. Where there are unmitigable risks of corrup-tion in procurement, outcome payment de-sign, or evaluation.

Understanding of the appropriate conditions for the use of this mechanism will continue to evolve as the field matures.



As of October 1, 2015, 49 SIBs are being utilized in developed countries to provide high-quali-

ty preschool education, reduce prison recidivism, avoid foster care placement, and increase youth employment, among other social issues. SIBs have thus far developed on three continents—Eu-rope, North America, and Australia (Figure 2).

The first SIB was implemented in 2010 in the United Kingdom for the purpose of reducing pris-on recidivism among short-term male prisoners.

This was followed by several more in the U.K. and by the first SIB outside of the U.K.—in the Unit-ed States—also for prison recidivism. The high-est number of impact bonds can be found in the U.K., which has 31 SIBs. The country with the sec-ond-highest number of SIBs is the United States, with eight as of October 1, 2015. Continental Eu-rope’s first SIBs were established in Germany and the Netherlands, followed by Belgium, Portugal, and Switzerland. Australia has implemented three SIBs, and Canada has developed one SIB.

2. THE GLOBAL LANDSCAPE OF IMPACT BONDS

SwitzerlandPeruIndiaPortugalCanadaBelgiumNetherlandsGermanyAustraliaU.S.U.K.

60

50

40

30

20

10

0

Mar

-10

May

-10

Jul-1

0Se

p-10

Nov-

10Ja

n-11

Mar

-11

May

-11

Jul-1

1Se

p-11

Nov-

11Ja

n-12

Mar

-12

May

-12

Jul-1

2Se

p-12

Nov-

12Ja

n-13

Mar

-13

May

-13

Jul-1

3Se

p-13

Nov-

13Ja

n-14

Mar

-14

May

-14

Jul-1

4Se

p-14

Nov-

14Ja

n-15

Mar

-15

May

-15

Jul-1

5Se

p-15

Figure 2. Impact bond contracts over time, October 2015

Source: Authors’ research.





In addition, two development impact bonds have been contracted, one in in Rajasthan, India, for girls’ education,8 and the second for cocoa and coffee production in Peru.9 Over 40 projects are underway in Latin America, Africa, the Middle East, South Asia, East Asia, and the Pacific Is-lands to establish impact bonds in a variety of sec-tors, including health and education.10

Thus far, impact bonds have been used for inter-ventions where inputs are complex but outcomes are relatively simple to measure. While many of the first SIBs were in the social issue area of crimi-nal justice, other areas have since gained traction. The 51 impact bonds contracted as of October 1, 2015 have outcomes in the sectors of employment, homelessness, child welfare, criminal justice, edu-cation, health, and agriculture (see Figure 3).11

Figure 3. Active impact bonds by sector,

October 2015

Employment19

Homelessness/Employment

7Homelessness

5

ChildWelfare

8

Criminal Justice4

Education4

Homelessness/Child Welfare

1

Health2

Agriculture1

Source: Authors research.

Eight impact bonds address the intersections of homelessness with child welfare or employment. Impact bonds may be particularly well suited to help individuals and families that need support

across a range of sectors, because they can unite service providers and government agencies around agreed upon outcomes. Further, impact bonds have not been used to finance tradition-al core government services, such as primary education or law enforcement, but rather have financed interventions such as per-primary edu-cation, mentoring for youth, and family therapy. As such, these interventions are often areas with histories of service provision by nongovernmental organizations.

This brief will present the policy implications of a recent Brookings Institution report on the 38 SIBs contracted as of March 1, 2015. A cursory exam-ination of the SIBs contracted between March and October 2015 leads us to conclude that the central findings from the landscape review of the first 38 transactions still apply at the time of publication of this brief. Apart from the two DIBs, all of the 11 new SIBs are in the same regions as the first 38, and nine of the 11 are in the same sectors, with the exception being two SIBs in the health sector. Below we draw out some of the main policy-rele-vant findings from our detailed review.

2.1 There is enormous flexibility in the way an impact bond can be structured and wide variation in the roles of the various stakeholders.

Impact bonds have been structured as one of two models—either as an individual transac-tion with one contract for outcomes agreed upon by all parties or as a fund with multiple contracts for government-issued outcomes. In an impact bond fund (see Box 1), prices are set for outcomes per participant and are paid on a month-ly, quarterly, or yearly basis. Many of the metrics of success in these deals are outputs (completion of an activity), rather than outcomes (measures of impact on the individual).

8 Instiglio (2015). 9 Finance Alliance for Sustainable Trade (2015). 10 As of October 1, 2015.11 Note that this pie chart could be divided up differently. We use the outcomes of focus as the differentiating factor.



Box 1: Impact Bond FundsAs of March 1, 2015 there were two active impact bond funds in the world, the Innovation Fund and the Fair Chance Fund, both in the U.K. The Innovation Fund aims to improve education and employment outcomes for youth and was launched in two rounds: the first six SIBs in April of 2012 and next four SIBs in November of 2012. The Fair Chance Fund, including seven SIBs aiming to improve hous-ing, education, and employment for homeless youth, was launched in December of 2014. Since this study’s cut-off, the U.K. launched the Youth Engagement Fund in mid-March of 2015, including four SIBs for youth education and employment.

To establish these impact bond funds, the commissioning government department (the Department for Work and Pensions in the case of the Innovation Fund and Youth Engagement Fund, and the De-partment for Communities and Local Government in the case of the Fair Chance Fund) produced a list of outcomes and prices it was willing to pay for each outcome, called a rate card. The list and prices were developed through extensive research on the cost savings of reduced remedial assistance, such as unemployment benefits, that each outcome will yield.

Table 4: Innovation Fund Round 1 Rate Card

OUTCOMEPAYMENT PER

INDIVIDUAL

Improved behavior at school (Measured by a letter from a teacher) £800Stop persistent truancy (absent for over 10% of school days per year) £1,300Achievement of First National Qualifications Framework (NQF) Level 2 qualification £2,200Achievement of First NQF Level 1 qualification £700Entry into first employment including a training element £2,600Entry into sustained employment £1,000Completion of first NQL Level 3 training/ vocational qualifications £3,300Successful completion of an ESOL course £1,200Entry into education at NQF level 4 £2,000

Source: Centre for Social Impact Bonds (2013). Note: Outcomes can only be claimed once per individual. Total payments per individual are capped at £8,200 and each SIB contract has a maximum payment cap across all individuals.

The commissioning department (the outcome funder) then committed a pool of funding to pay for outcomes. In the case of the Fair Chance Fund, the commissioning department’s funding was sup-plemented by funding from the U.K. Cabinet Office, and in the Youth Engagement Fund, funding was supplemented by the U.K. Cabinet Office and Ministry of Justice. Partnerships of service providers, investors and, in some cases, intermediaries were invited to bid for contracts within the fund, bidding at a discount to the rates in the rate card. Contracts were then awarded to bidders based on a number of factors, including the discount of their bid.

After service provision began, the service provider or intermediary may submit claims of outcomes, using various forms of administrative data, on a monthly or quarterly basis to the commissioner. The outcome payments are then reinvested, or “recycled,” to continue funding service provision. Capital recycling allows for lower upfront capital commitment than SIBs where outcome payments are not recycled.



Rigorous (experimental or quasi-experimen-tal) evaluations of the interventions in SIB deals are not always necessary for measuring impact and determining repayment. The type of evaluation method used to determine outcome payments depends on the outcome metrics being measured and the desire of the investors and out-come funders to have evidence as to the causality of the outcomes.12 For the SIBs with set payments for outcomes per individual, administrative data are often used to determine outcome payments. Examples of the administrative data include use of special education, placement in out-of-home care (residential or foster care), employment status, and incarceration. Validated administrative data was the most common evaluation method, used in 28 of 38 SIBs. In the SIBs where payments are based on the comparison of the program beneficiaries to other comparable groups, more complex evaluations are required. In six SIBs, outcomes are measured in comparison to a his-torical baseline. Concurrent control or comparison groups were used for comparison in eight SIBs, of which four used matched comparison groups—a quasi-experimental evaluation method—and four used a randomized control trial (RCT). Although the majority of SIBs used administrative data to determine investor payment, many of these SIBs are also running quasi-experimental evaluations or RCTs to evaluate program impact. Though these evaluations are not tied to investor payment, the findings will help improve future services and more accurately predict the savings associated with the intervention.

Layered capital structures are a useful option, but are not necessary. Thirteen of the 38 SIB deals have been financed by layered capital struc-tures, which may include subordinate investment, recoverable grants, non-recoverable grants, or in-vestment guarantees in addition to senior invest-ment (see Figure 4). Subordinate investment is

repaid after senior investment, making it more risky. Subordinate investment is often structured as an equity investment, while senior invest-ment is often structured as debt. Non-recover-able grants are not repaid, and investment guar-antees are triggered to pay investors only if the program is unsuccessful. The U.S., Australia, and the U.K have used layered capital structures for SIBs, including subordinate investment, grants, and guarantees. Layered structures can be a useful strategy to bring investors with a range of priorities to the table. Capital guarantees, howev-er, are not necessary for a successful SIB. Only six of the 38 SIBs guarantee capital protection of some percentage of senior investment, where the guarantee is provided by the service provider or a foundation.

Upfront capital commitment in the U.S. has tended to be higher than in the U.K., where capital recycling (see Box 1) is more common. Nevertheless, in the cases where this recycling of outcome funder payments occurs, the total amount that goes toward the intervention is of-ten much larger. The smallest amount of upfront capital commitment is in the SIB in Portugal, with $148,000, and the largest is in the SIB in Massa-chusetts, with $24.5 million in upfront capital com-mitment, though the vast majority of this is from grants (See Figure 4). The largest upfront capital commitment not including grants is the Child-Par-ent Center Pay for Success Initiative in in Chicago in the U.S. ($16.9 million).

Returns are variable and dependent on out-comes in all of the SIBs, as in an equity in-vestment, though some SIBs offer a much nar-rower range of returns, more similar to debt. In the U.S., deals have tended to be structured more like debt, while U.K. deals have tended to be structured more like equity. As all returns in the existing SIBs are variable depending on outcomes

12 Neither a causal relationship between the intervention and the outcomes nor an accurate comparison of efficacy between programs can be determined without randomized selection of beneficiaries and/or matched control groups, because outcomes can be affected by the selection of program beneficiaries or other external factors.



and each has its own risk profile, it is impossible to make direct comparisons in the returns to in-vestors across deals. However, all SIBs do estab-lish a maximum potential return for investors or a maximum amount of money that the outcome funder will pay, also known as a maximum con-tract value. The only deal with no maximum con-tract value is the It’s All About Me SIB for adoption in the U.K., where outcome funders can elect to join the program. The maximum average annual return of the investments in SIBs is at or below

market rate, with a few examples of above-market return in riskier, subordinate tranches.13 Of the 12 programs that have disclosed a maximum aver-age annual return,14 the eight are between 3 and 8 percent and four SIBs have maximum returns between 9 and 30 percent.15

SIB management is dependent on the man-date and capacity of the actors, which range immensely. Impact bonds have followed one of three management structures, which are dictated

Links 4 Life Programme, U.K.Aspire Gloucestershire, U.K.

3SC Capitalise Progrmme, U.K.Local Solutions, U.K.

Ambition East Midlands, U.K.Home Group, U.K.Your Chance, U.K.

T&T Innovation, U.K.Street Impact, U.K.ThinkForward, U.K.

Energise Innovation, U.K.Fusion Housing, U.K.

Rewriting Futures, U.K.Outcomes for Children Birmingham, U.K.

Manchester City Council Vulnerble Children, U.K.Triodos New Horizons, U.K.**

Nottingham Futures, U.K.**It’s All About Me (IAAM), U.K.

Advance Programme, U.K.Essex Family Therapy, U.K.

ONE Service, U.K.Junior Code Academy, Portugal

Duo for a Job, BelgiumSocial Impact Bond Rotterdam The Netherlands

Sweet Dreams Supported Living Project, CanadaThe Newpin Social Benefit Bond, Australia

Benevolent Society Social Benefit Bond, AustraliaPartnering for Family Success Program, U.S.

Utah High Quality Preschool Program, U.S.Increasing Employment and Improving Public Safety, U.S.

Juvenile Justice Pay for Success Initiative, U.S.The NYC ABLE Project for Incarcerated Youth, U.S.Child-Parent Center Pay for Success Initiatie, U.S.

Chronic Individual Homelessness Pay for Success Initiative, U.S.

0 5 10 15 20 25

USD millions

Senior Investment

Subordinate Investment

Recoverable Grants or Investment Guarantees

Non-Recoverable Grants

Figure 4. Upfront capital commitment in SIBs, as of March 1, 2015

13 In comparison, the MSCI World Index of capital in developed countries had an annualized gross return of 11.7 percent over the last 5 years and 6 percent over the last 10 years. MSCI (2015).

14 Three SIBs have disclosed a maximum internal rate of return (IRR), which range from 2 to 13 percent.15 The highest possible maximum return is for the subordinate investor class in the Benevolent Society Social Benefit Bond in

Australia.

Source: Authors’ research.Note: Upfront capital commitment not publicly available for Thames Reach Ace, Prevista, Living Balance, and Eleven Augsberg SIBs**Approximate



by whether the intermediary, investors, or service providers play the important roles of managing the deal and conducting performance management. Leadership to pull the transaction together varies as well, with governments exerting relatively high-er effort in impact bond funds and service provid-ers’ contributions varying greatly based on ca-pacity. This flexibility in role designation allows a variety of entities to play different roles depending upon their area of expertise and local conditions.

The number of service providers per contract rang-es from 1 to 11 and the number of investors ranges from 1 to 44. In the SIB with 44 investors, in New York State, Bank of America Merrill Lynch managed the investment of 44 individual investors. In the SIB around an intervention by the nonprofit The Benev-olent Society in Australia, investment was raised through the marketing of more traditional “bonds,” which are readily saleable at any point in the project. Across the SIBs, investors comprise a wide range of organizations—including foundations, communi-ty development financial institutions (in the U.S.), impact investors, and commercial banks—and they vary greatly in the balance of their interest in finan-cial and social return.

Service providers have also acted as equity inves-tors in four SIBs in the U.K., alongside external, fixed-interest loans with senior repayment priority. Similarly, service providers have the opportunity to earn success fees in three of the SIBs in the U.K. Innovation Fund and have the opportunity to earn success fees and deferred service fees in one of the SIBs for prison recidivism in the U.S. In the four family support SIBs in the U.K., service pro-viders receive payments from the outcome funder for each child that enters the program (for exam-ple, upon entry into a therapeutic foster care pro-gram), in addition to payments for each program outcome (for example, when that child is reunited with their family). The enrollment payments pro-vide capital for operating costs of the program for each participant. Investors receive an agreed-up-on percentage of the payments the service provid-er receives.

Depending on whether a government wishes or is able to engage, impact bonds can take the form of a SIB, where the government is the outcome funder, or a DIB where a third party takes on the role of outcome funder. Outcome funds may also comprise any combination of third party, national government, and local government funds. While only two DIBs have been contracted to date (though they were not contracted before our study cutoff date), a number of others are in develop-ment. This variation in the structure may allow for the initial introduction of this financing model to systems where domestic government outcome funds are insufficient.

Other forms of payment-by-results contracts can be incorporated into impact bond funds. In one of the 10 SIBs in the Innovation Fund in the U.K., the municipal government provides the upfront money for the service, provides the service, and is paid by the national government based on outcomes. This is a simple payment-by-results contract between the municipal and national governments. In anoth-er SIB in the U.K., the service provider assumes all risk and has a fixed-rate loan from an investor, essentially a payment-by-results contract between the provider and the government with an external loan. The It’s All About Me bond for adoption in the U.K. provides another example of innovative ways to use the SIB structure across government levels. In this bond, investors have committed a set amount, and the outcome funders (local gov-ernments known as Local Authorities) and service providers have the opportunity to join the SIB and receive upfront capital as adoption cases arise. These SIBs are excellent examples of the potential for adaptation of the model based on local context.

2.2 The four most important factors for contracting a SIB were measurable outcomes, evidence of intervention impact, government support, and dedication and collaboration of the stakeholders.

In the majority of the deals, there was adminis-trative data or a system set up for data collection



related to intervention impact. In every deal, either outcomes or outputs related to service delivery were identified. In many there was evidence that partic-ular service providers had the capacity to deliver those services and to achieve the outcomes in question. There was quite some variation, how-ever, in the robustness of that evidence, and this depended to some extent on the risk appetite of investors. Government support was crucial in ev-ery deal. There is a broad range of actions poli-cymakers at national and local levels can take to support the SIB ecosystem, which are summarized in our report. These actions include tax incentives for impact investment, government contributions to outcome funds, authorization to use impact bonds in existing spending areas, public funding for fea-sibility studies and intermediaries, and support for knowledge-sharing networks. Finally, the stake-holders overwhelmingly highlighted the importance of all of the actors’ willingness to put in the nec-essary time and effort to develop these deals. In-termediaries, in particular, were required to spend

a great deal of time arranging many SIBs, but in other cases service providers, investors, and other nonprofits have led the contracting process when they have sufficient capacity to do so. For example, in Utah, the advocacy organization Voices for Utah Children had staff with a background in finance and developed the initial proposal for the structure of the SIB. SIB contracting has required philanthropic support for the time and effort spent on developing the deals, though actors are increasingly looking for ways to build these costs into the contracts.

2.3 Five of the 10 most common claims about the benefits of the impact bond mechanism are supported by stakeholder experiences in the first 38 SIBs.

The most important claim surrounding SIBs is that impact bonds lead to a shift in focus to outcomes. We find that the first 38 SIBs have truly transformed the conversation among participating government stakeholders about procurement of social services,

6. FOSTER INNOVATION IN DELIVERY

9. BUILD A CULTURE OF MONITORING+EVALUATION

8. INCENTIVIZE COLLABORATION

1. CROWD-IN PRIVATE FUNDING

7. DRIVE PERFORMANCE MANAGEMENT 5. ACHIEVE SCALE

2. INVEST IN PREVENTION

3. REDUCE RISK FOR GOVERNMENT

4. FOCUS ON OUTCOMES

10. SUSTAIN IMPACT

Figure 5. 10 common claims about impact bonds



transparency, and accountability. In essence, in-stead of paying for services, government pays for outcomes. At the same time, SIBs push service pro-viders to deliver on these outcomes. A second very important and related claim is that impact bonds drive performance management. Bringing private sector mentality into the provision of services (which often means getting government bureaucracies out) can lead to more efficient and effective delivery of social services. This has been mainly seen through the push toward outcome achievement and fidelity to the service delivery model and less in terms of adaptation of service provision along the way. Third, in the existing deals, this mechanism has held up to the claim that it stimulates collaboration. Fourth, if impact bond deals lead to larger systematic change, such as development of strong monitoring and eval-uation systems, that in itself would be an enormous contribution toward improving service delivery and, ultimately, many people’s lives. Finally, impact bonds can shift the focus of government away from cura-tive or remedial services and toward preventive ser-vices. These five trends could have huge economic implications for government and society, though it is

too soon to tell if what we observe within the SIB will materialize at a system level.

2.4 The first 38 impact bonds are not achieving substantial scale in absolute terms or fostering significant innovation, and we have yet to see whether impact bonds will crowd-in private funding, reduce risk for government, or sustain impact.

Of the 38 deals, 25 serve populations equal to or less than 1,000 individuals (see Figure 6). However, many of the deals had very specific target populations, so in relative terms the programs were serving an important part of that target population in a given setting. Despite the small scale in these 38 bonds, this is an area where impact bonds have evolved since the report’s cutoff—the Ways to Wellness SIB and the Educate Girls DIB, launched in late March 2015, aim to reach 11,000 and 18,000 ben-eficiaries, respectively. Impact bond funds could be utilized in the future to facilitate programs at greater scale as they establish a framework for scale across multiple organizations.

10,000

4,500

4,000

3,500

3,000

2,500

2,000

1,500

1,000

500

0

Cana

daPo

rtuga

lU.

K.U.

K.Ge

rman

yU.

K.U.

K.U.

K.Ne

ther

land

sU.

K.Be

lgiu

mU.

K.U.

K.U.

S.U.

K.U.

K.U.

K.U.

K.U.

K.Au

stra

liaU.

K.U.

K.U.

K.U.

S.U.

S.U.

K.U.

K.Au

stra

liaU.

K.U.

S.U.

S.U.

K.U.

K.U.

K.U.

S.U.

K.U.

S.

Num

ber o

f Tar

get B

enef

icia

ries

Figure 6. SIB target beneficiaries, as of March 1, 2015

Source: Authors’ research.Note: The SIB intending to reach 10,000 beneficiaries in Rikars Island Prison in New York City closed one year early in July 2015.



Few of the programs financed by the first 38 SIBs were truly innovative in that they had never been tested before, but many were innovative in that they applied interventions in new settings or in new combinations. If the risk of service ineffec-tiveness of these interventions is indeed relative-ly low, the SIB mechanism does not significantly reduce risk for government. However, the recent results of the SIB for prison recidivism in Rik-ers Island Prison in New York City demonstrate that the risk shift has the potential to be quite meaningful—the intervention, one of the most in-novative to be funded by a SIB, did not have a statistically significant impact, and thus there were no outcome payments from government to inves-tors.16 A number of factors prevented the program from producing the expected results, including is-sues with mixing between treatment and control groups, the Department of Education’s decision not to provide ancillary support, and a decrease in other funds available to the service provider.17 Our research also shows mixed evidence on the power of impact bonds to crowd-in private funding. If a program is successful, government ultimately pays for the program and investor money is re-turned. The impact bond does allow government to pay at the end of the service—when savings have been realized in some cases—rather than the beginning, which may allow for programs that would not have been possible within the govern-ment budget at the start of the program. There is some evidence that traditional service funders are spending more and new actors are investing, but we have yet to see whether this money will be reused in the social sector if the investment is returned. Finally, five years since the first impact bond, we have yet to see whether impact bonds will lead to sustained impact on the lives of ben-eficiaries beyond the impact bond contract dura-tion by stimulating sustained government support or by instilling a culture of outcome achievement, monitoring, and evaluation.

2.5 Impact bonds have the potential to contribute to the improvement of social service delivery, though thus far the deals have been complex and costly in time and expertise.

Deal development has proven to be challenging due to the steep learning curve of this new form of collaboration involving many different types of play-ers and the complexity of multiyear contracts and legal constraints. The need to determine outcome metrics and conduct complex budget analytics and calculate the costs and benefits of interventions adds to the difficulties in putting a deal together.

Beyond the investment amount for the service pro-vision are additional costs to the transaction that include intermediary services and technical assis-tance, evaluation, and legal fees. These amounts can be structured either within the deal or outside the deal. From our review of the landscape, it is apparent that the first impact bonds have been time-intensive and costly operations. Much of the initial work was done on a pro bono basis, and some of it has been designed with success pay-ments to be made to intermediaries if the deal is successful in achieving outcomes. As the impact bond market grows, some developments could reduce these costs or at least make them more sustainable. First, there will be some reduction in transaction costs that result from learning and stan-dardization of processes. Nevertheless, for new ac-tors in new settings, adaptation of these processes will be necessary. Second, some intermediaries are beginning to conduct initial scoping exercises for impact bond transactions in which they build in the costs for their time. Third, the development of philanthropic and government support for scoping and feasibility exercises will help to make this work possible for actors wanting to engage. A SIB de-velopment fund has recently been established, for example, for the development of a portfolio of three

16 Anderson & Phillips (2015).17 Porter (2015).



SIBs in Utah. This “collaborative financing” model brings together public, philanthropic, and private capital to support the critical development phase for the SIB transactions.18

Finally, the development of innovation funds or rate cards (see discussion above), which allow for larger investments over multiple service providers or for the achievement of multiple outcomes, is another way to reduce relative transactions costs.

2.6 Results from the interventions financed by the first two impact bonds have been mixed, but more positive intervention results are being reported.

So far, few deals have completed at least one set of payments to investors. In the ONE Service SIB in Peterborough U.K., interim payments are made for each of three cohorts if the percentage differ-ence in prison recidivism between the treatment and control groups is greater than 10 percent, or a payment is made at the end of the program if the percentage difference between the treatment and control group exceeds 7.5 percent across all of the three cohorts. To date, one of the cohorts has reached evaluation, achieving an impressive reduction in recidivism of 8.4 percent relative to a comparable national baseline. This percentage change does not meet the interim payment target, but is on track for the overall payment at the end

of the program.19 As noted previously, the SIB for criminal justice in New York City did not result in a statistically significant reduction in recidivism by year three of the intervention, leading to a loss of $1.2 million for the senior investor, Goldman Sachs, and a loss of $6 million for the loan guar-antor, Bloomberg Philanthropies.20

Of the 10 total SIBs in the Innovation Fund, at least two have already fully repaid investors the principal and will pay the additional outcomes pay-ments through the rest of the year as a return on the investment.21 The Newpin Social Benefit Bond in Australia has disclosed that investors received a 7.5 percent interest payment in the first year of the program and an 8.9 percent interest payment in the second, which is in the middle of the range of interest that investors can receive.22 The results from the first year of the Benevolent Society Social Benefit Bond project a 5 percent return to senior investors and an 8 percent return to subordinate investors when payments are made at the end of the contract term.23 Finally, in SIB for preschool services in Utah, 109 of the 110 students deter-mined to be at-risk for use of remedial education in the first cohort of the program did not require remedial services, resulting in a payment to the senior investors of $267,000, which is 95 percent of the savings accrued to the state for special ed-ucation avoidance.24

18 Interview with Liya Schuster, Third Sector Capital. May 14, 2015. 19 Social Finance U.K. (2014).20 Anderson & Phillips (2015). 21 Social Finance U.K. (2015).22 Social Ventures Australia (2015).23 The Benevolent Society (2014).24 Wood (2015).



3. CONCLUSIONS

Prospects are bright for the implementation of more impact bonds (or some derivative of them) worldwide, though this will take con-certed effort on the part of many stakeholders. Currently, there are over 60 SIBs in development in high-income countries and over 30 in develop-ment in low- and middle-income countries. Given the complexity of deal construction, the evidence and outcome identification necessary to even be-gin to develop a deal, and the willingness of out-come funders to enter into these contracts, bring-ing these and more impact bonds to fruition will require a multi-pronged approach. The challeng-es are likely to be much larger in low- and mid-dle-income country contexts. First, support from philanthropy will continue to be crucial to the field; these contributions have played a key role in the building of the impact bond ecosystem to date. For example, they can help by funding the collec-tion of data and evidence necessary to start the conversation about outcome-based financing with outcome funders. Second, legislation and policy action on the part of government will also be im-portant in the future development of the market. Third, technical assistance or advising govern-ments will be very important to help ensure that the needs of beneficiary populations are taken into account and that the costs and benefits make sense. This will be even more important as new sectors are explored for deal development. Final-ly, expertise will be crucial in the areas of financial modeling, contracting, performance management, and in conducting due diligence of nonprofits.

It is very likely that the impact model develop-ment process, structure, and application will continue to be adapted in the future. Thus far we have seen SIBs developed in fields with a com-plex set of inputs but with simple outcomes. It is likely that there will be more impact bonds devel-oped in these same types of sectors but that future impact bonds will come to include, for example, a wider range of interventions in early childhood de-velopment (maternal and child health, parenting, and child welfare), health (particularly in preven-tive care), housing, and water and sanitation. The types of interventions within these sectors that are most probable include services that cater to par-ticularly underserved or marginalized populations as well as to ones that provide improvements in the margin to existing services, such as in quality improvements when access is not an issue. Im-pact bonds could also be used more experimen-tally where investors are interested, for instance, in testing innovative ideas for service provision or outcome funders would like to test which interven-tions or service providers deliver services most effectively. While risky, these propositions could represent high future value for both investors and outcome funders.

Can impact bonds tackle some of the large-scale social issues facing the world today, in particular in developing country contexts? Impact bonds alone will not be able to finance all required services at scale in developing country contexts given the outsize sums of private capi-tal that would be needed to fund these services,





the availability and capacity of service providers to provide at large scale, and the availability of outcome funders to make payments to investors if outcomes are achieved. One way to address some of these issues, as mentioned above, is through the impact bond funds. In developing country contexts, such funds could be supported by multiple large donor agencies, for example, but the problem of capacity may remain. At the same time, enormous scale may not be the purpose of impact bonds. If smaller deals are able to shift how governments and service providers think about service provision and to consider contracting de-livery to non-state actors when appropriate and if outcomes become the main focus, this could have important ripple effects on service delivery more broadly. For this reason, it is imperative that gov-ernment be involved in contract design and that stakeholders make a strong effort to convince a government agency to contribute to outcome pay-ments. Working solely with third-party outcome funders in a DIB may facilitate negotiations and may provide an important demonstration effect; however, the systemic change resulting from the use of the mechanism is likely to be much less significant than if the government were involved,

as in a SIB. This argument will be familiar to those who worry about government ownership of aid programs.

As the impact bond market grows, we expect that some of the challenges faced in the first five years of deal development will be reduced. The main challenge will be the complexity of the deal and the time and costs involved in designing contracts. Already, stakeholders are beginning to think cre-atively about how to go about this. As discussed above, this will require the contribution of all in-volved, and it will take some creative thinking and collective effort.

To move forward there must be increased trans-parency and knowledge sharing. The develop-ment of communities of practice, workshops, con-ferences, and easily accessible online content can foster this movement. More research about how this nascent field develops will be critical to cap-ture lessons learned, contextualize them within the bigger picture of social sector financing and service provision, and apply them to real-world problems with the populations in need at the cen-ter of the discussion.

Please refer to the main report “The Potential and Limitations of Impact Bonds: Lessons Learned from the First Five Years of Experience” for a list of study participants, fact sheets on the 38 impact bonds contract-ed as of March 1, 2015, and an inventory of legislation supporting impact bonds.



REFERENCESAnderson, J. and A. Phillips. 2015. “What we

Learned from the Nation’s First Social Impact Bond.” Huffington Post. (http://www.huffingtonpost.com/james-anderson/what-we-learned-from-the-_1_b_7710272.html [July 10,2015]).

The Benevolent Society. 2014. “Social Benefit Bond: Investor Report for the nine months ended 30 June 2014.” The Benevolent Society. (http://www.benevolent.org.au/~/media/Benevolent/About/BenevolentSoci-ety_SBB_Investor_Report_2014%20pdf.ashx [October 1, 2015]).

Bloomgarden, D., M. Eddy, and Z. Levey. 2014. Social Impact Bonds and Education in Latin America. GEMS Education Solutions.

Burand, D. 2013. “Globalizing Social Finance: How Social Impact Bonds and Social Im-pact Performance Guarantees Can Scale Development.” NYU Journal of Law and Business 9.

Center for Global Development and Social Fi-nance. 2013. Investing in Social Outcomes: Development Impact Bonds. Center for Global Development and Social Finance.

Centre for Social Impact Bonds. 2013. “De-partment for Work and Pensions Inno-vation Fund,” U.K. Cabinet Office, (data.gov.uk/sib_knowledge_box/depart-ment-work-and-pensions-innovation-fund).

Finance Alliance for Sustainable Trade. 2015. “Cocoa and Coffee Production by Indigenous Ashaninka People.” Finance Alliance for Sus-tainable Trade. (http://www.fastinternational.org/en/node/2208 [September 25, 2015]).

Instiglio. 2015. “Girls Education India.” Instiglio. (http://www.instiglio.org/en/girls-educa-tion-india/ [September 25, 2015]).

Jones, N., Jones, H., Steer, L., and Datta, A. 2009. “Improving impact evaluation produc-tion and use,” Working Paper 300, Overseas Development Institute.

MSCI. 2015. “MSCI World Index (USD).” MSCI. (https://www.msci.com/resources/fact-sheets/index_fact_sheet/msci-world-index.pdf [September 25, 2015]).

Porter, Eduardo. 2015. “Wall St. Money Meets So-cial Policy at Rikers Island.” Economic Scene. The New York Times. July 28, 2015. (http://www.nytimes.com/2015/07/29/business/econ-omy/wall-st-money-meets-social-policy-at-rik-ers-island.html?_r=0 [July 28, 2015]).

PricewaterhouseCoopers. 2014. “Asset Manage-ment 2020: A Brave New World.” Pricewa-terhouseCoopers. (https://www.pwc.com/gx/en/asset-management/publications/pdfs/pwc-asset-management-2020-a-brave-new-world-final.pdf [September 25, 2015]).

Saltuk, Y. 2014. Spotlight on the Market: the Impact Investor Survey. J.P. Morgan and the Global Impact Investing Network.

Social Finance U.K. 2014. “Peterborough Social Impact Bond Reduces Reoffending by 8.4%; Investors on Course for Payment in 2016.” Social Finance U.K. (www.socialfinance.org.uk/wp-content/uploads/2014/08/Peter-borough-First-Cohort-Results.pdf [May 10, 2015]).

Social Finance U.K. 2015. “First Social Impact Bonds to Return Investor Capital.” Social Finance U.K. (http://www.socialfinance.org.uk/wp-content/uploads/2015/07/Press-re-lease-First-Social-Impact-Bonds-to-return-investor-capital.pdf [October 1, 2015]).

Social Ventures Australia. 2015. “Newpin Social Benefit Bond: Annual Investor Report.” So-cial Ventures Australia. (http://www.dpc.nsw.gov.au/__data/assets/pdf_file/0009/173547/Newpin_Investor_Report_2015.pdf [July 10, 2015]).

Wood, Benjamin. 2015. “Preschool paying off for Goldman Sachs and Utah kids according to United Way.” The Salt Lake Tribune. (http://www.sltrib.com/home/3032598-155/pre-school-paying-off-for-goldman-sachs [Octo-ber 7, 2015]).

http://www.huffingtonpost.com/james-anderson/what-we-learned-from-the-_1_b_7710272.html



http://www.benevolent.org.au/~/media/Benevolent/About/BenevolentSociety_SBB_Investor_Report_2014%20pdf.ashx




http://data.gov.uk/sib_knowledge_box/department-work-and-pensions-innovation-fund



http://www.fastinternational.org/en/node/2208

http://www.fastinternational.org/en/node/2208

http://www.instiglio.org/en/girls-education-india/

http://www.instiglio.org/en/girls-education-india/

https://www.msci.com/resources/factsheets/index_fact_sheet/msci-world-index.pdf



http://www.nytimes.com/2015/07/29/business/economy/wall-st-money-meets-social-policy-at-rikers-island.html?_r=0




https://www.pwc.com/gx/en/asset-management/publications/pdfs/pwc-asset-management-2020-a-brave-new-world-final.pdf




http://www.socialfinance.org.uk/wp-content/uploads/2014/08/Peterborough-First-Cohort-Results.pdf



http://www.socialfinance.org.uk/wp-content/uploads/2015/07/Press-release-First-Social-Impact-Bonds-to-return-investor-capital.pdf




http://www.dpc.nsw.gov.au/__data/assets/pdf_file/0009/173547/Newpin_Investor_Report_2015.pdf



http://www.sltrib.com/home/3032598-155/preschool-paying-off-for-goldman-sachs



1775 Massachusetts Ave., NWWashington, D.C. 20036

brookings.edu

Documents

POLICY RECOMMENDATIONS FOR THE APPLICATIONS OF IMPACT … · 2016-07-21 · Policy Recommendations for the Applications of Impact Bonds: A Summary of Lessons Learned from the First